Upload existing datasets to Zenodo and move docs

[namsaraeva]

Currently all of the files (images) needed for the example notebooks are saved in the repository. Moreover, the heavy autophagy checkpoints are also saved automatically. This should be changed to "on demand" behaviour- we need to upload those files to Zenodo and write corresponding dataloader functions.

[sophiamaedler]

closing as already implemented